Speculative Lock Region Bug (Low Priority)

[dz333]

Currently, since lock regions aren't managed the same way as normal locks,
if a speculative thread is killed and holds a lock region, that region will become deadlocked.

We just need to add some metadata to the stage graph to mark the stages where a lock region is held to add the appropriate release statements to the kill rules.

E.g.:

start(rf);
---
spec_barrier(); //if the thread is killed here, the rf lock region will be deadlocked
---
end(rf);

This is only relevant for multi-cycle lock regions and so isn't a high priority bug (they're hardly used and we may deprecate them in favor of a different system in the future).